Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

Knowledge Graph Error Detection with Contrastive Confidence Adaption

Authors: Xiangyu Liu, Yang Liu, Wei Hu

AAAI 2024 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results demonstrate that CCA outperforms state-of-the-art baselines, especially in detecting semantically-similar noise and adversarial noise.
Researcher Affiliation Academia 1 State Key Laboratory for Novel Software Technology, Nanjing University, China 2 National Institute of Healthcare Data Science, Nanjing University, China {xyl, yliu20}.EMAIL, EMAIL
Pseudocode No The paper describes the model architecture and components but does not provide structured pseudocode or algorithm blocks.
Open Source Code Yes Datasets and source code are available at https://github.com/nju-websoft/CCA.
Open Datasets Yes We conduct our experiments on FB15K-237 (Toutanova et al. 2015) and WN18RR (Dettmers et al. 2018).
Dataset Splits No The paper discusses training and testing, and mentions 'We randomly divide it into training and testing sets, Dtrain and Dtest' for adversarial noise generation, but does not specify validation splits for the main experiments or general train/test/validation splits.
Hardware Specification Yes All experiments are conducted on two Intel Xeon Gold 6326 CPUs, 512GB RAM, and one NVIDIA RTX A6000 GPU.
Software Dependencies No We leverage the BERT-base model from huggingface as the PLM. We use Py Torch to implement our model and employ the Adam W optimizer and a cosine decay scheduler with a linear warm-up for optimization. (No specific version numbers are provided for PyTorch or huggingface modules).
Experiment Setup No The grid search is used for hyperparameter tuning. (No specific hyperparameter values like learning rate, batch size, or number of epochs are explicitly stated for the final model setup in the main text).